The normalization is needed to have one single scale for the data that you obtained, the alignment also it´s necessary for this task, and all depends on the nature of your experiment, in the metabolomics case you can take one Quality Control to divide all the samples between this sample (taking also your QC), this can give you new recalculated values for all your features with a new scale in relationship with this QC value (where you must have a representative pool of all your samples). Here is a link from the software which i used for my metabolomic data, but exist a lot of free software to do this depending on the format of your data.

http://www.nonlinear.com/support/progenesis/lc-ms/faq/v4.1/how-normalisation-works.aspx

Best

Josué

Depends very much on the what kind analysis or specific area of -omics research you are doing. You might find these basic article helpful understanding the utilization of such data mining techniques.

1. Why, How and When to Scale your Features

https://medium.com/greyatom/why-how-and-when-to-scale-your-features-4b30ab09db5e

2. Normalization vs Standardization — Quantitative analysis

https://towardsdatascience.com/normalization-vs-standardization-quantitative-analysis-a91e8a79cebf

3, The great importance of normalization of LC–MS data for highly‐accurate non‐targeted metabolomics:

Article The great importance of normalization of LC-MS data for high...

4. Normalization Method:

https://www.sciencedirect.com/topics/computer-science/normalization-method

Dear Tim Young,

Regarding your suggestions, I would like to make my questions in more details.

1. In my metabolomics experiement, I did not use the internal standard, hence, I cannot use the internal standard based normalization method to eliminate the confounding factors.

2. To be more specific, I have used the LC-MS based metabolomics to find significant metabolites between two groups (my biological sample is plasma). In the case of QC based normalization method, I do not know in which case this method is considered as the best choice in normalization.

3. My aim is to minimize the non-biological variation prior to the statistically analysis step to make metabolite profiles comparable in comparative samples, leading to determining potential bio markers in a particular disease.

I hope that based on the update of my questions, you may consider giving me more suggestions and tips.

Thank you so much,

Pham Quynh Khoa.

Hi Pham,

you do not have any internal standard, nor did you measure additional QC samples. This limits your options of which method to apply. Your main sources of non-biological variation will be batch and run order effects (maybe also sample processing).

I assume that you analyze a small sample set where all samples have been measured in one single batch (than you can neglect this one).

I hope that you measured at least 6 replicates of each of your two groups (A and B) and that you randomized the measurement order, i.e. that you did not measure all A samples and than all B samples but measured A B A B ... instead. If you did the latter you can remove potential runorder effects using an ANOVA based method (10.1007/978-1-4939-7819-9_20). You may have further factors you can control for (are your replicates from independent experiments or can you further group them? did you process your samples all together or in smaller groups?).

If you measured all A followed by all B you can not distinguish between technical and biological effects and should try to measure again.

Bests, Jan

Relative qualitative data analsis is still possible with

the aim to identify peobable biomarkers if not absolute quantitave analysis without IC standards.

As Lisec suggested, samples replicates, randomization of samples before data acquisition, and also standardized sample collection and preprocessing, instrument sensitivity and specific, inter and intra day instrument variation among others should be considered.

Dear Jan Lisec and Khalid Muzaffar,

In my experiment, my sample set was quite large (around 100 study samples including the two groups (50-50). Of course, I randomized the run order of study samples between the two groups. In addition, each of the QC samples, which were prepared from a pool sample of around 100 study samples, has been analyzed every 8 study sample analytical set.

On the other hand, all blank, QC and study samples have been processed together by a person and then, all analytical extracts of these samples have been put in the auto-sampler at the same time. The run time for each sample was around 45 minutes and the total for this batch was 1 week.

May you please show me based on what criteria to select the best sample normalization method for a given metabolomic data? In other hand, when we examine a given metabolomic data, which parameters will determine an optimal sample normalization method for a given metabolomic data.

Many thanks,

Pham Quynh Khoa.

Dear Tim Young,

I greatly appreciate your detailed suggestion. In fact, I followed all the above steps which you mentioned and now, I have the more meaningful data after the blank subtraction, missing value filling (kNN), and removal of some features that show %RSD greater than 30% in QC samples. I think that the next step for my data pre-treatment is the normalization and scaling.

As far as I know, these steps are quite challenging because I should select the best methods for my data. However, I do not have much knowledge in the normalization and scaling, namely which one is the best choice, how to do it and what each method means.

If convenient for you, may you please help me explain each method and send me your excel sheet with your formulas, an example of your own data and so on via my email [email protected]?

Thank you so much,

Best regards,

Pham Quynh Khoa.

Dear Tim Young,

Thank you so much for your kindness and detailed explanation.

I will apply these methods to my data and hope that they can fix the variation.

I will let you know the goodness of my data when I finish treating it.

Thank you so much again,

Best regards,

Pham Quynh Khoa.

Hi,

This is a very interesting and relevant question for the entire research community working in metabolomics.

Please also feel free to register (for free to create a login with your Email ID!) and join the International Metabolomics Society’s Forum: http://www.metabolomics-forum.com/index.php?action=forum;?action=forum#c12 to ask more specific questions any software, database, approach and platform (mass spectrometry, NMR) issues where you can get very specific responses from expert users and researchers!

Thanks again,

Biswa

Metabolism , that is mean , the degradation of molecules (carbohydrates,lipids and proteins into small molecules are called intermediates and carbon dioxide and water and energy these process are called Catabolism,and another processes are called anabolism these are containing several pathways for synthesis big molecules such as (carbohydrates,lipids and fat, with expended energy which are liberated from catabolism pathways

Hi Pham Quynh Khoa,

I found the following article very useful.

Misra BB. Data normalization strategies in metabolomics: Current challenges, approaches, and tools. European Journal of Mass Spectrometry. 2020;26(3):165-174. doi:10.1177/1469066720918446

Do have a read.

Hope this helps.

jaime